Node Configuration with step wise list box population for end user

I am trying to build a node that has multi step configuration points. There are a lot of use cases for this, but I will use a simplified example to demonstrate what I am trying to do.

Consider a setup where you have several databases, 1 production, and 4 different testing databases for different types of testing. Also, assume that not all testing environments contain all clients that are available in production. Further, we want to control what queries a user can run in production to prevent long running queries designed for testing from being run when the user is working with a production instance.

In this example, I want to be able to let the user select a database instance, then, based on the selected instance, return a list of available clients in that instance. The user then selects an appropriate client. Once all the information is gathered, the workflow returns a list of prepared queries that the end user can consume downstream in their flow.

I know that this would result in the end user having to wait after making a choice in each step for the select box in the next step to populate with appropriate values, but this is not an issue.

the flow above works as expected. I can make appropriate choices at appropriate times, but when I wind it up into a node, I either can’t see the settings at all, or I can’t get them to populate with out actually running the node after making a selection at point 1.

Has anybody managed to successfully create a step wise configurable node?

Hi @jmyers and welcome to the forum.

I think what you want to do here is achievable - it sounds like you would benefit from wrapping some of your nodes above into a few components, to give your users interaction points for selecting databases, clients, and related queries. Maybe this short blog post will help orient you:

https://www.knime.com/blog/metanode-or-component

(If you are ultimately looking to run this workflow on KNIME Server using the webportal, you will want to use wigdet nodes instead of configuration nodes, but the setup is much the same for both.)

Also, if you really want to dig into detail on component setup, our documentation is here:

https://docs.knime.com/latest/analytics_platform_components_guide/index.html

1 Like

Thanks @ScottF

To set up one configurable point like this is easy. I can do that. I’m trying to have a single control with one config screen in a KNIME workflow that allows the user to do this at 1 spot with out having to run bits of the flow and stop to make selections downstream.

So, on the config screen User will select database, wait a second or 2 for the next drop-down for available clients to populate and then select client.

Remember, this is a contrived example. My real life example will get more complex. You can think of the over all config as being a workflow where each time the flow reaches a point that requires a selection that will impacts a downstream choice, then the config screen waits for the user to make a selection, and runs the next series of steps in the flow …

Ah, OK, I understand now. As I mentioned above, if you run this workflow on KNIME Server via the Webportal, that’s exactly how things would work - the user makes a selection, the next bit of the workflow runs until another interaction is needed - and so on.

In KNIME Analytics Platform, I’m afraid you’re going to have to run the individual parts of the workflow in series.

1 Like

Thanks @ScottF. Not the answer I was hoping for, but also not one that I was not expecting. Makes sense though.

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.